Variations of Stable Isotopes in Daily Precipitation in a Monsoon Region

Abstract

The stable isotopes of hydrogen and oxygen in precipitation provide a useful reference for the study of hydrological processes, but concerns have been raised regarding the established patterns in their variations in a monsoon climate zone. In this study, stable isotopes (δ¹⁸O and δ²H) of 539 daily precipitation samples from seven hydrometeorological stations in the Beijing area were used to investigate short-term isotopic variations and the controlling factors at a region with a monsoon climate. The δ¹⁸O in the precipitation increases from the northwest to the south, which is controlled by the monsoon from the south and continental moisture from the northwest. Consistently, both the high δ¹⁸O values and low deuterium excess values from May to September reveal the impact of the monsoon. The amount effect is significant during the monsoon season. In contrast, the T effect is significant, with a gradient of 0.4‰ per °C during the non-monsoon season. A Rayleigh distillation model indicates that the moisture source and residual vapor fraction are the two most important factors in controlling the δ¹⁸O patterns in precipitation in a monsoon region, independent of temperature. The result of this study is helpful for the understanding of the regional atmospheric water cycle.

1. Introduction

Stable isotope records of paleo-precipitation preserved in the natural archives (e.g., speleothems, tree rings, ice cores, corals, and lake/ocean sediments) have provided information on past climate change at the regional and global scales [1,2,3,4] with the assumption that the δ¹⁸O of the archives (δ¹⁸O_a) follows the changes of precipitation δ¹⁸O (δ¹⁸O_p) [5,6,7]. However, the climatic significance of δ¹⁸O_a remains a debate. For example, depleted δ¹⁸O in groundwater (δ¹⁸O _g) was reported for the late Pleistocene groundwater in North China and West Africa, which was attributed to temperature changes [8,9,10,11] or monsoon strength [12]. The enriched stalagmite δ¹⁸O (δ¹⁸O_s) from monsoon regions was observed in the late Pleistocene, as a proxy of the local rainfall amount [13,14,15], or variations in monsoon intensity [3,16,17,18], or changes in moisture sources [19,20,21,22,23]. Furthermore, some studies emphasized that δ¹⁸O_s and δ¹⁸O_g are biased to the heavy rain events during the wet season [24,25,26,27,28,29].

Resolving these debates requires a good understanding of the isotopic composition of precipitation. Numerous studies have investigated precipitation isotopes, especially the relationship between T or P and isotopes [30,31,32]. Araguas-Araguas et al. (1998) [33] analyzed the stable isotope composition of precipitation over Southeast Asia and concluded that δ¹⁸O_P values increase with T in the north and decrease with P in the south. Subsequently, Johnson and Ingram (2004) [34] demonstrated that the multiple regression method should be used rather than the traditional single regression method with only T or P. Li et al. (2014) [31] further detailed a strong positive relationship between δ¹⁸O_P and T in winter, and a negative linear correlation between δ¹⁸O_P and summer P. Tan (2014) [35] found that δ¹⁸O_P reflects neither the amount effect nor strength of a single summer monsoon, but the intensity contrast between two summer monsoons: the southwest and southeast monsoons in the monsoon regions of China. The stable isotope systematics are quite complex in the monsoon-affected regions of China, and thus further work should be completed with higher resolution data. In recent years, event precipitation and even intra-event sampling for isotopic analysis have been carried out in order to gain insight into the complex processes/factors controlling precipitation isotopes [36,37,38]. Li et al. (2015) [39] identified different moisture sources dominating the variations in precipitation isotopes during an extreme event in Beijing. Although a great deal of work is related to stable isotopes in precipitation, little attention has been paid to the contributions of non-monsoon precipitation in the monsoon margin area.

Here, we argue that high-resolution data from both the temporal and spatial sampling of precipitation events can provide more information on the factors and processes controlling the variations in modern precipitation isotopes. The Beijing region, the northern extent of the East Asian summer monsoon (EASM) in northern China, was selected as a study area. We established a daily precipitation event sampling network with a spatial resolution of one station every 2500 km², covering the mountains and plains over the Beijing region. Based on the observation throughout 2015, the primary objectives of this study include: (1) Do moisture sources, moisture trajectories, and/or local metrological conditions control the isotopic variations of precipitation? (2) What is the relationship between precipitation isotope and T or P at the eventful, seasonal, and annual scales?

2. Data and Methods

2.1. Study Area

Beijing, the capital of China, is in the northern rim of the North China Plain (Figure 1a). Due to the typical continental monsoon climate, there are four distinct seasons in Beijing. The non-monsoon season is cold and dry, due to northerly winds blowing from high-latitude areas, while the monsoon season is hot and wet because of the east and south airflow carrying moisture from the southern Pacific Ocean and Indian Ocean [40,41]. The annual precipitation is 574 mm, of which 70% occurs during the monsoonal season from May to September (Figure 1b,c). The average annual temperature is approximately 12 °C. The mean air temperature in the mountain area is lower (10.2 °C at YQ station) than that in the plains (15.1 °C at BTC station).

2.2. Sampling and Analysis

To better characterize the moisture circulation over the Beijing area, we established the Beijing Network of Isotopes in Precipitation (BNIP) based on the principal hydrologic stations of the Beijing Water Authority in May 2014 (Figure 1). A rain collector composed of a polyethylene bottle and funnel was placed outside at each station, with a ping-pong ball placed at the funnel mouth to prevent evaporation. Snow samples were collected using a pail installed on the ground. After each snowfall event, snow samples were melted at room temperature. The precipitation samples were collected at 8:00 am in the next day after the precipitation events. Samples were immediately sealed after collection in 20 mL fully filled bottles to avoid evaporation after rainfall, and then transported to the laboratory and stored at 4 °C until the analysis. Stable isotopes were analyzed on a Picarro L1102-i Laser Absorption Water Isotope Spectrometer at the Water Isotope and Water–Rock Interaction Laboratory, Institute of Geology and Geophysics, Chinese Academy of Sciences. Results were reported as δ¹⁸O (δ¹⁸O = (R_sample/R_standard − 1) × 1000) using the Vienna Standard Mean Ocean Water (VSMOW) standard. The analytical precisions for δ²H and δ¹⁸O were 0.5‰ and 0.1‰, respectively. Meteorological variables, including precipitation amount (P, mm), temperature (T, °C), and relative humidity (RH, %) were observed after event precipitation sampling at each observation station. Due to the limitations imposed by field conditions, seven stations with data integrity in 2015 (Figure 1) were selected for analysis in this study. Deuterium excess (d-excess), a second-order parameter that combines both oxygen and hydrogen isotopic species, is defined by the following expression (Dansgaard, 1964): d-excess = δ²H − 8 × δ¹⁸O.

2.3. Moisture Trajectories

To deduce the probable moisture trajectories in the study area, we generated backward trajectories using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model, developed by the National Oceanic and Atmospheric Administration (NOAA) Air Resources Laboratory [42]. Three-dimensional wind fields for trajectory calculation were derived from the Global Data Assimilation System (GDAS) of the National Centers for Environmental Prediction (NCEP). The most rainfall in summer over Beijing precipitates from clouds with base heights of 1000–2000 m [43]. Therefore, we used an average height of 1500 m in the HYSPLIT model to obtain the moisture trajectories.

2.4. Spatial Interpolation

Spatial interpolation provides a method for estimating the isotopic compositions of precipitation where data are not available, and a smoothed trend surface is generated that captures the geographic data variability. In this study, the contours of δ¹⁸O and δ²H values for precipitation over Beijing were mapped using Kriging interpolation in ArcGIS 10.8 software. The surrounding GNIP stations, including Shijiazhuang, Tianjin, Jinzhou, and Changchun, were incorporated to eliminate the boundary effect. We calculated the amount-weighted δ¹⁸O from June to August as summer, and from December to February as winter.

3. Results

3.1. Spatial Distribution of Annual Mean Stable Isotopes in Precipitation

In the observed stations, mean annual δ¹⁸O, δ²H, and deuterium excess values ranged from −7.6 to −9.6‰, −51.9 to −65.0‰, and 9.3~11.5‰, respectively (see

Table 1. BNIP station locations, annual precipitation amount, annual mean temperature, and measured annual mean precipitation isotopic values.

Station	Longitude (°E)	Latitude (°N)	Elevation (m)	δ2H (‰)	δ18O (‰)	dexcess (‰)	P (mm)	T (°C)
YQY	117.375	40.637	477	−57.0	−8.5	11.2	540.8	12.3 *
ZJF	116.782	40.614	186	−59.4	−8.6	9.3	596.1	14.2 *
YQ	115.993	40.463	492	−65.0	−9.6	11.5	531.0	10.2
YC	115.894	40.033	238	−51.9	−7.7	9.3	423.7	15.3 *
FHY	116.694	39.615	11	−58.0	−8.7	11.4	474.1	16.8 *
ZF	115.688	39.576	108	−49.7	−7.6	11.4	514.9	13.5 *
BTC	116.379	39.976	54	−59.7	−8.8	10.3	600.3	15.1

Note. * denotes the average temperature of all the recorded data at the sampling events.

). Although the isotopic values were relatively stable, a clear pattern was found: the spatial distribution of δ¹⁸O and δ²H was almost parallel to the latitude lines (Figure 2a,b), with increasingly negative values at higher (more northern) latitudes, which is in good agreement with the δ¹⁸O precipitation contour map for China [31]. In contrast, the deuterium excess was almost the same over the whole Beijing region.

The spatial variations in δ¹⁸O of precipitation during the summer and winter seasons are shown in Figure 2c,d. During the winter season, the precipitation contours indicated a clear increasing trend from northwest to south. During the summer season, precipitation δ¹⁸O was more enriched, but within a narrow range. We also found that the summer δ¹⁸O values were close to the annual values, likely because precipitation is concentrated in the summer (Figure 1a,c).

3.2. Seasonal Variations in Precipitation Isotopes

Seasonal changes in δ¹⁸O were similar at all stations (Figure 3a), with a broad maximum, −2.0~−9.5‰ and −19.2~−67.7‰ for δ¹⁸O and δ²H values, respectively, during the summer months. The minima, −9.1‰~−21.5‰ and −157.6~−45.8‰ for δ¹⁸O and δ²H values, respectively, occurred in the winter months. Such a seasonal variation was also observed by Wen et al. (2010) [40] for both vapor and precipitation isotopes at a station in Beijing, and by Yamanaka et al. (2004) [30] for multi-year precipitation isotopes in the northern North China Plain.

Different from the stable isotopes, the annual variations in deuterium excess show a dominant ‘V’-shaped pattern in Beijing (Figure 3b). The arrival of the monsoon at these stations was marked by an increase in rainfall (Figure 1b,c), and a corresponding increase (drop) in δ¹⁸O (deuterium excess) values. The much more pronounced minima for δ¹⁸O were observed during the winter months, associated with relatively high d-excess values. These elevated d-excess values indicate the presence of continental air masses producing precipitation during the winter season.

3.3. Local Meteoric Water Line

Isotopic ratios of D/H and ¹⁸O/¹⁶O in precipitation are closely related, lying on a meteoric water line, which provides isotopic input functions for hydrological studies. The local meteoric water line (LMWL) for Beijing (Figure 4), derived from amount-weighted monthly values in this study, was calculated as:

δ²H = 7.5δ¹⁸O + 8.2 (R² = 0.97, n = 77)

(1)

The slope and intercept of the LMWL increased continuously with increasing P (

Table 2. Correlation equations for δ²H and δ¹⁸O values of individual precipitation samples with different precipitation amount ranges in Beijing, China.

Precipitation AmountRange	δ2H/δ18O Correlation Equations	R2	N
≥0	δ2H = 6.7δ18O − 2.4	0.93	539
≥1.0	δ2H = 7.0δ18O + 1.1	0.94	400
≥5.0	δ2H = 7.4δ18O + 6.4	0.96	199
≥10.0	δ2H = 7.7δ18O + 8.9	0.96	115

). For all the rain events, the slope of the δ²H/δ¹⁸O relationship was 6.7, within the range calculated using a numerical secondary evaporation model [44]. When the individual P was >5 mm, the LMWL based on individual data had a slope close to that based on the monthly weighted data. For the storm events, e.g., ≥10mm, the slope of the δ²H/δ¹⁸O relationship was obviously higher at 7.7. Secondary evaporation is always observed for light rain events, or at the beginning of storms in an arid environment [45]. Based on our data from Beijing, the change to precipitation isotopes due to secondary evaporation was most pronounced for rain samples with P < 0.5 mm (slope = 6.2).

4. Discussion

4.1. Relationship between δ¹⁸O and T or P

The regression equation between δ¹⁸O and T or P (the so-called T or P effect of Dansgaard (1964) [46]) was calculated based on both daily and monthly data. For monthly data, a clear relationship between the mean monthly δ¹⁸O and T during the non-monsoonal season is shown in Figure 5a, with a high correlation coefficient (R²) of 0.74. The δ¹⁸O/T slope was 0.4 ‰/°C, which is in good agreement with that of the typical westerlies domain (Urumqi station in Xinjiang Uygur Autonomous Region as an example). Points from the monsoonal season were plotted below the regression line for T and δ¹⁸O. In contrast, the δ¹⁸O during the monsoonal season showed a significant P effect (Figure 5b); the monthly slope was approximately −2.9‰/100 mm (R² = 0.53). This agrees with the conclusions of Li et al. (2014) [31] that T and P effects dominate in different seasons.

To demonstrate and interpret the relationship more clearly, daily data from BTC and YQ stations were selected for further evaluation. For daily data, δ¹⁸O only correlated well with T when T was below 16 °C, and the gradient of the T effect at the two stations was similar, with a value of 0.4‰/°C. When T was above 16 °C, the T effect disappeared, as shown by the scattered points indicated with a rectangle in Figure 5c. It should be noted that conditions with T > 16 °C only occur in the monsoon season.

We compared our findings with the results from the Gaoshan and Houxia stations in the Tianshan Mountains (Figure 5c). For all four stations, the T effect only exists within a certain temperature range, below 0 °C for Gaoshan and Houxia stations, and below 16 °C for YQ and BTC stations. However, when there is no T effect, the δ¹⁸O values change very little at Gaoshan and Houxia stations, while they fluctuate remarkably over the Beijing region. The Tianshan Mountains are dominated by westerlies throughout the year, and sub-cloud evaporation and moisture recycling processes alter the δ¹⁸O values when T is above 0 °C [47,48,49,50]. The moisture in Beijing is controlled by the westerlies and polar air masses in winter (Figure 6). Similar to other regions with prevailing westerlies, the temperature effect is significant in Beijing during the non-monsoon seasons, as shown in Figure 5a,c. During the monsoon seasons, rain samples reveal a depleted isotope signature despite generally higher summer temperatures. Because δ¹⁸O is affected by the sub-cloud evaporation process when P is small, the heavy daily rainfall events (>5 mm/day) from YQ and BTC stations are clustered. A significant inverse relationship between δ¹⁸O and P is observed (Figure 5d). When the moisture sources at BTC station are limited to from the south and west (northwest) based on HYSPLIT trajectories, R² increases to 0.53 and 0.49, respectively, illustrating more significant correlations, which also imply the significant effect of rainfall sources on precipitation δ¹⁸O to some extent.

4.2. Moisture Sources over the Beijing Region

The moisture sources affecting Beijing are composed of both continental moisture from the northwest and marine moisture from the south. However, their contributions to the Beijing vapor each month are quite different. The HYSPLIT model was used to show moisture trajectories for the study period over the Beijing region. During the winter season, the moisture was primarily sourced from the northwest; during the summer season, more moisture was transported from the south and east. When we selected rainfall events with P > 20 mm to illustrate their trajectories, most were sourced from the south and east (Figure 6).

The isotopes incorporated in continental moisture from the northwest are generally more depleted in heavy isotopes and have higher deuterium excess values than those in marine moisture from the southwest [33]). This is consistent with the spatial patterns illustrated by our data in Figure 2, and the seasonal pattern illustrated in Figure 3. To understand the effect of moisture sources on the precipitation isotopes, a Rayleigh distillation model [7] was employed based on continuously selected daily data from 16–20 July 2015.

Figure 7 displays the isotopic variations during rainfall events at seven stations. A decreasing isotopic trend was observed during the first four days, in good agreement with the Rayleigh distillation model. However, an increasing trend for isotopes was observed on 20 July 2015 with the addition of a new air mass. The increasing isotopic compositions were first found at BTC, ZF, and YC stations, indicating a southwestern moisture source. This is in agreement with the HYSLTIP results shown in Figure 8. Moisture extended from offshore areas westward over Beijing during the first four days, while air from further inland contributed more to rainfall on 20 July 2015.

Initially, the moisture derived from offshore areas (initial moisture) arrived in Beijing from east to west (Figure 8b). The δ¹⁸O value of the initial moisture could be obtained from the precipitation δ¹⁸O and corresponding P using an inverse analysis in the Rayleigh fractionation model. For this purpose, we assumed there was no extra moisture added; thus, the initial isotopic values can be calculated using the following two equations:

$${\delta }^{18}{\mathrm{O}}_{rain(f)}\approx {\delta }^{18}{\mathrm{O}}_{(v)}+{\epsilon }^{18}{\mathrm{O}}_{l-v}\approx {\delta }_o^{18}{\mathrm{O}}_v+{\epsilon }^{18}{\mathrm{O}}_{l-v}\times \ln f+{\epsilon }^{18}{\mathrm{O}}_{l-v}$$

(2)

$$f=1-\frac{\mathrm{P}}{{\mathrm{P}}_{\mathrm{sum}}}$$

(3)

where ${\epsilon }^{18}{\mathrm{O}}_{l-v}$ is 9.3‰ assuming that the temperature is ~25 °C. Using P and δ¹⁸O values for the first two days, the initial isotopic values were calculated, as shown in Figure 9a. Increasingly negative values at higher (more westward) longitudes were obtained with a good correlation coefficient of 0.88 to which is in good agreement with the meteorological results from the HYSPLIT models. The initial δ¹⁸O value for the YQY station was about −0.6‰, similar to that of seawater [51] which provided further evidence for an eastern moisture source from offshore. There was no remarkable difference in isotopes on 19 July, and the decreasing isotopic trend (Figure 9b) from south to north for the rainfall on 20 July was primarily caused by the southern moisture sources, with a negligible P effect. New southern air masses, mainly derived from terrestrial evaporation, were more enriched compared to isotopes of residual moisture. Therefore, precipitation δ¹⁸O was mainly controlled by the initial isotopic values, residual vapor fraction, and input of new air masses. The initial isotopic values were again dominated by moisture source and rainout process before arriving at the rainfall location, and the input of new air masses should be seen as a moisture source factor. Therefore, moisture sources and residual vapor fraction during the rainout process dominated the rhythm of δ¹⁸O in the monsoon-affected region of Beijing.

5. Conclusions

Daily precipitation samples were collected throughout 2015 and analyzed from seven stations covering the mountain and basin region of Beijing. The dominant factors controlling δ¹⁸O_p and changes in δ¹⁸O_p related to temperature and precipitation variabilities in the monsoon-affected region were identified.

The annual δ¹⁸O_p and δ²H_p show a clear spatial pattern that increases from south (east) to north (west), while the deuterium excess varies little. The summer δ¹⁸O_p is relatively stable, while the winter δ¹⁸O_p increases from northwest to south. Such a spatial distribution feature is mainly controlled by northwest continental and south marine moisture sources. The seasonal changes in δ¹⁸O are very similar at all stations, with high values in summer and low values in winter. Both the high δ¹⁸O values and low deuterium excess values from May to September reveal the impact of the monsoon. The local meteoric water line (LMWL) for Beijing, derived from amount-weighted monthly values is δ²H = 7.5 δ¹⁸O + 8.2 (R² = 0.97, n = 77). The smaller slope of 6.7 derived from individual events is attributed to a local secondary evaporation process. A significant T effect is found, with a gradient of 0.4‰/°C when T is below 16 °C. In the monsoon season, when T is always above 16 °C, the P effect dominates, particularly when P > 5 mm.

A Rayleigh distillation model was employed to verify the factors controlling the precipitation isotopes based on selected continuous event data for 16–20 July 2015. We found that the moisture sources and residual vapor fraction are the most important factors controlling the rhythm of δ¹⁸O. The limited observed period from these stations is recognized, and an extension of such isotopic studies over a longer period, and the inclusion of more stations will obtain more quantitative determination of the various effects on the formation of precipitation in the monsoon-affected region.